aoiandroid/TranslateBlue-v2-GGUF

TranslateBlue v2 (GGUF Q4_K_M)

Translation model focused on 29 languages with emphasis on African languages, in GGUF format for use with llama.cpp and compatible runtimes (iOS, Android, desktop).

Model description

• Base model: Qwen3-4B-Instruct
• Format: GGUF, quantized with Q4_K_M
• Size: ~2.3 GB
• Training: LoRA fine-tuning on parallel translation data (10,000 steps, 16 LoRA layers)
• Training data: 563,986 sentence pairs from 29 languages

Intended use

• Text translation between the supported languages, especially to/from African languages
• Offline translation in mobile and desktop apps (e.g. TranslateBlue)
• Batch or interactive translation via llama.cpp or bindings (Swift, Python, etc.)

Supported languages (29)

Code	Language	Code	Language	Code	Language
sw	Swahili	ha	Hausa	yo	Yoruba
ig	Igbo	am	Amharic	zu	Zulu
xh	Xhosa	af	Afrikaans	so	Somali
rw	Kinyarwanda	sn	Shona	tw	Twi
ee	Ewe	wo	Wolof	ny	Chichewa
ti	Tigrinya	nso	Northern Sotho	tn	Tswana
om	Oromo	ve	Venda	nd	Ndebele
ar	Arabic	fr	French	pt	Portuguese
es	Spanish	de	German	zh	Chinese
ja	Japanese	ko	Korean	en	English

Limitations

• Best for short to medium sentences; very long texts may lose quality.
• Low-resource pairs may be less accurate than high-resource ones.
• No built-in language detection; source and target languages should be specified in the prompt.

How to use

Prompt format

Use a clear translation instruction, for example:

Translate from English to Swahili:

Hello, how are you?

Or:

Translate from French to Hausa:

Bonjour, comment allez-vous?

With llama.cpp (command line)

llama-cli -m translateblue-v2-q4_k_m.gguf \
  -p "Translate from English to Swahili:\n\nHello, how are you?" \
  -n 64 --temp 0.3

With Python (llama-cpp-python)

from llama_cpp import Llama
llm = Llama(model_path="translateblue-v2-q4_k_m.gguf")
out = llm("Translate from English to Swahili:\n\nHello, how are you?", max_tokens=64, temperature=0.3)
print(out["choices"][0]["text"])

In iOS / Swift (e.g. TranslateBlue)

The model is registered as TranslateBlue v2 (GGUF). After downloading via the app, it runs with the built-in LlamaCppService using the same prompt format above.

Training details

Setting	Value
Base model	Qwen3-4B-Instruct
Method	LoRA
LoRA layers	16
Steps	10,000
Training samples	563,986
Validation loss	~2.5

License

MIT.

Citation

If you use this model in research or a product, please cite the base model (Qwen3) and the TranslateBlue project as appropriate.